media sentiment system

[nutmix]

1. Scrape phrases/words from twitter, telegram and possibly others in the future
2. manually editable list of twitter hashtags, telegram groups etc.
3. Manually editable db of phrases, with templates, e.g. "is shit" is bad, "the shit" is good. Possibly template driven.
4. Look for resources online to do this (e.g. algorithms, databases of phrases)
5. Provide per day results per coin.
6. Provide graphs of trends for given time ranges.
7. no need to go below 1 hour time resolutions.
8. for twitter, examples are #BTC, #cyrpto #ETC etc.
9. for telegram, examples are ICOCountdown.
10. should be written in go.
11. private repo
12. english only

Beautiful soup was mentioned as a parser, but this is geared towards html

[gz-c]

if we're doing language detection we should track this to get a sense of activity in different markets